This study advances a dual-analytical framework examining the heterogeneous associations between social media use and university students' academic adjustment within the Chinese cultural context, integrating Social Cognitive Theory and the Differential Susceptibility to Media Effects Model. Through a mixed-methods approach combining variable-centered moderated mediation and person-centered latent profile analysis (N = 4,216), we identify two critical mechanisms: (1) social anxiety substantially mediates the relationship between social media use and academic adjustment, accounting for a significant proportion of this association, and (2) social self-efficacy moderates this statistical mediation pathway. Person-centered analyses reveal four distinct digital engagement configurations: high-intensity high-efficacy (36%), moderate-high intensity high-efficacy (28%), moderate-low intensity high-efficacy (20%), and low-intensity low-efficacy (16%). Profile-specific statistical mediation analyses demonstrate differential associations across profiles, with the strongest indirect relationship in the low-intensity low-efficacy group and the weakest in high-intensity high-efficacy users. These findings challenge uniform-effect assumptions by demonstrating how psychological resources co-occur with usage intensity to create distinctive associative patterns. While cross-sectional data preclude causal inference, results suggest theoretical and practical implications for understanding configuration-specific relationships between social media engagement and academic functioning. This dual-analytical framework advances theoretical integration of psychological mechanisms and configurational patterns in digital learning environments, potentially informing future longitudinal investigations of adaptation processes in educational technologies.